Remove parse_xml_records function

* Remove parse_xml_records function from helpers module and corresponding unit test
* Replace all calls of that function with the appropriate class method

tests/conftest.py

@@ -5,8 +5,8 @@
 
 import transmogrifier.models as timdex
 from transmogrifier.config import SOURCES, load_external_config
-from transmogrifier.helpers import parse_xml_records
 from transmogrifier.sources.datacite import Datacite
+from transmogrifier.sources.transformer import XmlTransformer
 
 
 @pytest.fixture(autouse=True)
@@ -48,12 +48,14 @@ def runner():
 
 @pytest.fixture()
 def datacite_records():
-    return parse_xml_records("tests/fixtures/datacite/datacite_records.xml")
+    return XmlTransformer.parse_source_file(
+        "tests/fixtures/datacite/datacite_records.xml"
+    )
 
 
 @pytest.fixture()
 def datacite_record_all_fields():
-    source_records = parse_xml_records(
+    source_records = XmlTransformer.parse_source_file(
         "tests/fixtures/datacite/datacite_record_all_fields.xml"
     )
     return Datacite("cool-repo", source_records)
@@ -71,7 +73,7 @@ def marc_content_type_crosswalk():
 
 @pytest.fixture()
 def oai_pmh_records():
-    return parse_xml_records("tests/fixtures/oai_pmh_records.xml")
+    return XmlTransformer.parse_source_file("tests/fixtures/oai_pmh_records.xml")
 
 
 @pytest.fixture()

tests/test_datacite.py

@@ -1,4 +1,3 @@
-from transmogrifier.helpers import parse_xml_records
 from transmogrifier.models import (
     AlternateTitle,
     Contributor,
@@ -170,7 +169,7 @@ def test_datacite_transform_with_all_fields_transforms_correctly(
 
 
 def test_datacite_transform_missing_required_datacite_fields_logs_warning(caplog):
-    source_records = parse_xml_records(
+    source_records = Datacite.parse_source_file(
         "tests/fixtures/datacite/datacite_record_missing_datacite_required_fields.xml"
     )
     output_records = Datacite("cool-repo", source_records)
@@ -195,7 +194,7 @@ def test_datacite_transform_missing_required_datacite_fields_logs_warning(caplog
 
 
 def test_datacite_transform_with_optional_fields_blank_transforms_correctly():
-    source_records = parse_xml_records(
+    source_records = Datacite.parse_source_file(
         "tests/fixtures/datacite/datacite_record_optional_fields_blank.xml"
     )
     output_records = Datacite("cool-repo", source_records)
@@ -218,7 +217,7 @@ def test_datacite_transform_with_optional_fields_blank_transforms_correctly():
 
 
 def test_datacite_transform_with_optional_fields_missing_transforms_correctly():
-    source_records = parse_xml_records(
+    source_records = Datacite.parse_source_file(
         "tests/fixtures/datacite/datacite_record_optional_fields_missing.xml"
     )
     output_records = Datacite("cool-repo", source_records)
@@ -241,7 +240,7 @@ def test_datacite_transform_with_optional_fields_missing_transforms_correctly():
 
 
 def test_datacite_with_attribute_and_subfield_variations_transforms_correctly():
-    source_records = parse_xml_records(
+    source_records = Datacite.parse_source_file(
         "tests/fixtures/datacite/datacite_record_attribute_and_subfield_variations.xml"
     )
     output_records = Datacite("cool-repo", source_records)

tests/test_dspace_dim.py

@@ -1,10 +1,9 @@
 import transmogrifier.models as timdex
-from transmogrifier.helpers import parse_xml_records
 from transmogrifier.sources.dspace_dim import DspaceDim
 
 
 def test_dspace_dim_transform_with_all_fields_transforms_correctly():
-    source_records = parse_xml_records(
+    source_records = DspaceDim.parse_source_file(
         "tests/fixtures/dspace/dspace_dim_record_all_fields.xml"
     )
     output_records = DspaceDim("cool-repo", source_records)
@@ -133,7 +132,7 @@ def test_dspace_dim_transform_with_all_fields_transforms_correctly():
 
 
 def test_dspace_dim_transform_with_attribute_variations_transforms_correctly():
-    source_records = parse_xml_records(
+    source_records = DspaceDim.parse_source_file(
         "tests/fixtures/dspace/dspace_dim_record_attribute_variations.xml"
     )
     output_records = DspaceDim("cool-repo", source_records)
@@ -188,7 +187,7 @@ def test_dspace_dim_transform_with_attribute_variations_transforms_correctly():
 
 
 def test_dspace_dim_transform_with_optional_fields_blank_transforms_correctly():
-    source_records = parse_xml_records(
+    source_records = DspaceDim.parse_source_file(
         "tests/fixtures/dspace/dspace_dim_record_optional_fields_blank.xml"
     )
     output_records = DspaceDim("cool-repo", source_records)
@@ -204,7 +203,7 @@ def test_dspace_dim_transform_with_optional_fields_blank_transforms_correctly():
 
 
 def test_dspace_dim_transform_with_optional_fields_missing_transforms_correctly():
-    source_records = parse_xml_records(
+    source_records = DspaceDim.parse_source_file(
         "tests/fixtures/dspace/dspace_dim_record_optional_fields_missing.xml"
     )
     output_records = DspaceDim("cool-repo", source_records)

tests/test_dspace_mets.py

@@ -1,10 +1,9 @@
 import transmogrifier.models as timdex
-from transmogrifier.helpers import parse_xml_records
 from transmogrifier.sources.dspace_mets import DspaceMets
 
 
 def test_dspace_mets_transform_with_missing_optional_fields_transforms_correctly():
-    dspace_xml_records = parse_xml_records(
+    dspace_xml_records = DspaceMets.parse_source_file(
         "tests/fixtures/dspace/dspace_mets_record_optional_fields_missing.xml"
     )
     output_records = DspaceMets("dspace", dspace_xml_records)
@@ -20,7 +19,7 @@ def test_dspace_mets_transform_with_missing_optional_fields_transforms_correctly
 
 
 def test_dspace_mets_transform_with_blank_optional_fields_transforms_correctly():
-    dspace_xml_records = parse_xml_records(
+    dspace_xml_records = DspaceMets.parse_source_file(
         "tests/fixtures/dspace/dspace_mets_record_optional_fields_blank.xml"
     )
     output_records = DspaceMets("dspace", dspace_xml_records)
@@ -36,7 +35,7 @@ def test_dspace_mets_transform_with_blank_optional_fields_transforms_correctly()
 
 
 def test_dspace_mets_with_attribute_and_subfield_variations_transforms_correctly():
-    dspace_xml_records = parse_xml_records(
+    dspace_xml_records = DspaceMets.parse_source_file(
         "tests/fixtures/dspace/dspace_mets_record_attribute_and_subfield_variations.xml"
     )
     output_records = DspaceMets("dspace", dspace_xml_records)
@@ -85,7 +84,7 @@ def test_dspace_mets_with_attribute_and_subfield_variations_transforms_correctly
 
 
 def test_dspace_mets_transform_with_all_fields_transforms_correctly():
-    dspace_xml_records = parse_xml_records(
+    dspace_xml_records = DspaceMets.parse_source_file(
         "tests/fixtures/dspace/dspace_mets_record_all_fields.xml"
     )
     output_records = DspaceMets("dspace", dspace_xml_records)

tests/test_ead.py

@@ -1,12 +1,13 @@
 import logging
 
 import transmogrifier.models as timdex
-from transmogrifier.helpers import parse_xml_records
 from transmogrifier.sources.ead import Ead
 
 
 def test_ead_record_all_fields_transform_correctly():
-    ead_xml_records = parse_xml_records("tests/fixtures/ead/ead_record_all_fields.xml")
+    ead_xml_records = Ead.parse_source_file(
+        "tests/fixtures/ead/ead_record_all_fields.xml"
+    )
     output_records = Ead("aspace", ead_xml_records)
     assert next(output_records) == timdex.TimdexRecord(
         source="MIT ArchivesSpace",
@@ -216,7 +217,7 @@ def test_ead_record_all_fields_transform_correctly():
 
 
 def test_ead_record_with_missing_archdesc_logs_error(caplog):
-    ead_xml_records = parse_xml_records(
+    ead_xml_records = Ead.parse_source_file(
         "tests/fixtures/ead/ead_record_missing_archdesc.xml"
     )
     output_records = Ead("aspace", ead_xml_records)
@@ -230,7 +231,7 @@ def test_ead_record_with_missing_archdesc_logs_error(caplog):
 
 
 def test_ead_record_with_missing_archdesc_did_logs_error(caplog):
-    ead_xml_records = parse_xml_records(
+    ead_xml_records = Ead.parse_source_file(
         "tests/fixtures/ead/ead_record_missing_archdesc_did.xml"
     )
     output_records = Ead("aspace", ead_xml_records)
@@ -244,7 +245,7 @@ def test_ead_record_with_missing_archdesc_did_logs_error(caplog):
 
 
 def test_ead_record_with_attribute_and_subfield_variations_transforms_correctly():
-    ead_xml_records = parse_xml_records(
+    ead_xml_records = Ead.parse_source_file(
         "tests/fixtures/ead/ead_record_attribute_and_subfield_variations.xml"
     )
     output_records = Ead("aspace", ead_xml_records)
@@ -470,7 +471,7 @@ def test_ead_record_with_attribute_and_subfield_variations_transforms_correctly(
 
 
 def test_ead_record_with_blank_optional_fields_transforms_correctly():
-    ead_xml_records = parse_xml_records(
+    ead_xml_records = Ead.parse_source_file(
         "tests/fixtures/ead/ead_record_blank_optional_fields.xml"
     )
     output_records = Ead("aspace", ead_xml_records)
@@ -488,7 +489,7 @@ def test_ead_record_with_blank_optional_fields_transforms_correctly():
 
 
 def test_ead_record_invalid_date_and_date_range_are_omitted(caplog):
-    ead_xml_records = parse_xml_records(
+    ead_xml_records = Ead.parse_source_file(
         "tests/fixtures/ead/ead_record_attribute_and_subfield_variations.xml"
     )
     output_record = next(Ead("aspace", ead_xml_records))
@@ -507,7 +508,7 @@ def test_ead_record_invalid_date_and_date_range_are_omitted(caplog):
 
 
 def test_ead_record_correct_identifiers_from_multiple_unitid(caplog):
-    ead_xml_records = parse_xml_records(
+    ead_xml_records = Ead.parse_source_file(
         "tests/fixtures/ead/ead_record_attribute_and_subfield_variations.xml"
     )
     output_record = next(Ead("aspace", ead_xml_records))
@@ -516,7 +517,7 @@ def test_ead_record_correct_identifiers_from_multiple_unitid(caplog):
 
 
 def test_ead_record_with_missing_optional_fields_transforms_correctly():
-    ead_xml_records = parse_xml_records(
+    ead_xml_records = Ead.parse_source_file(
         "tests/fixtures/ead/ead_record_missing_optional_fields.xml"
     )
     output_records = Ead("aspace", ead_xml_records)

tests/test_helpers.py

@@ -4,7 +4,6 @@
 from transmogrifier.helpers import (
     generate_citation,
     parse_date_from_string,
-    parse_xml_records,
     validate_date,
     validate_date_range,
 )
@@ -189,11 +188,6 @@ def test_generate_citation_with_all_fields():
     )
 
 
-def test_parse_xml_records_returns_record_iterator():
-    records = parse_xml_records("tests/fixtures/datacite/datacite_records.xml")
-    assert len(list(records)) == 38
-
-
 def test_parse_date_from_string_success():
     for date in [
         "1930",

tests/test_marc.py

@@ -3,12 +3,11 @@
 from bs4 import BeautifulSoup
 
 import transmogrifier.models as timdex
-from transmogrifier.helpers import parse_xml_records
 from transmogrifier.sources.marc import Marc
 
 
 def test_marc_record_all_fields_transform_correctly():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_all_fields.xml"
     )
     output_records = Marc("alma", marc_xml_records)
@@ -489,7 +488,7 @@ def test_marc_record_all_fields_transform_correctly():
 
 
 def test_marc_record_attribute_and_subfield_variations_transforms_correctly():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_attribute_and_subfield_variations.xml",
     )
     output_records = Marc("alma", marc_xml_records)
@@ -706,7 +705,7 @@ def test_marc_record_attribute_and_subfield_variations_transforms_correctly():
 
 
 def test_marc_record_with_blank_optional_fields_transforms_correctly():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_blank_optional_fields.xml"
     )
     output_records = Marc("alma", marc_xml_records)
@@ -727,7 +726,7 @@ def test_marc_record_with_blank_optional_fields_transforms_correctly():
 
 
 def test_marc_record_with_missing_optional_fields_transforms_correctly():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_missing_optional_fields.xml"
     )
     output_records = Marc("alma", marc_xml_records)
@@ -746,7 +745,7 @@ def test_marc_record_with_missing_optional_fields_transforms_correctly():
 
 
 def test_marc_record_missing_leader_logs_error(caplog):
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_missing_leader.xml"
     )
     output_records = Marc("alma", marc_xml_records)
@@ -760,7 +759,7 @@ def test_marc_record_missing_leader_logs_error(caplog):
 
 
 def test_marc_record_missing_008_logs_error(caplog):
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_missing_008.xml"
     )
     output_records = Marc("alma", marc_xml_records)
@@ -774,7 +773,7 @@ def test_marc_record_missing_008_logs_error(caplog):
 
 
 def test_create_subfield_value_list_from_datafield_with_values():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_all_fields.xml"
     )
     datafield = next(marc_xml_records).find_all("datafield", tag="130")[0]
@@ -786,15 +785,15 @@ def test_create_subfield_value_list_from_datafield_with_values():
 
 
 def test_create_subfield_value_list_from_datafield_with_blank_values():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_blank_optional_fields.xml"
     )
     datafield = next(marc_xml_records).find_all("datafield", tag="130")[0]
     assert Marc.create_subfield_value_list_from_datafield(datafield, "ad") == []
 
 
 def test_create_subfield_value_string_from_datafield_with_values():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_all_fields.xml"
     )
     datafield = next(marc_xml_records).find_all("datafield", tag="130")[0]
@@ -805,7 +804,7 @@ def test_create_subfield_value_string_from_datafield_with_values():
 
 
 def test_create_subfield_value_string_from_datafield_with_blank_values():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_blank_optional_fields.xml"
     )
     datafield = next(marc_xml_records).find_all("datafield", tag="130")[0]
@@ -894,7 +893,7 @@ def test_loc_crosswalk_code_to_name_returns_name(caplog, loc_country_crosswalk):
 
 
 def test_get_main_titles_record_with_title():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_all_fields.xml"
     )
     assert Marc.get_main_titles(next(marc_xml_records)) == [
@@ -903,31 +902,35 @@ def test_get_main_titles_record_with_title():
 
 
 def test_get_main_titles_record_with_blank_title():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_blank_optional_fields.xml"
     )
     assert Marc.get_main_titles(next(marc_xml_records)) == []
 
 
 def test_get_main_titles_record_without_title():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_missing_optional_fields.xml"
     )
     assert Marc.get_main_titles(next(marc_xml_records)) == []
 
 
 def test_get_source_record_id():
-    marc_xml_records = parse_xml_records(
+    marc_xml_records = Marc.parse_source_file(
         "tests/fixtures/marc/marc_record_all_fields.xml"
     )
     assert Marc.get_source_record_id(next(marc_xml_records)) == "990027185640106761"
 
 
 def test_record_is_deleted_returns_true_if_deleted():
-    deleted_record = parse_xml_records("tests/fixtures/marc/marc_record_deleted.xml")
+    deleted_record = Marc.parse_source_file(
+        "tests/fixtures/marc/marc_record_deleted.xml"
+    )
     assert Marc.record_is_deleted(next(deleted_record)) is True
 
 
 def test_record_is_deleted_returns_false_if_not_deleted():
-    marc_record = parse_xml_records("tests/fixtures/marc/marc_record_all_fields.xml")
+    marc_record = Marc.parse_source_file(
+        "tests/fixtures/marc/marc_record_all_fields.xml"
+    )
     assert Marc.record_is_deleted(next(marc_record)) is False